http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
설치류 후각 뇌의 혈역학적 반응을 이용한 합성곱 신경망 기반 냄새 분류
김준민(Jun-Min Kim),이우람(Wooram Lee) 한국정보기술학회 2021 한국정보기술학회논문지 Vol.19 No.7
Although various odor detection and identification devices are being developed, such as air pollution, sewage pollution, fire, and explosive detection, there are still many difficulties in reproducing the mammalian sense of smell. Research is being actively conducted based on the fact that neural response signals in the olfactory brain of mammals appear differently depending on the odorant. In particular, near-infrared spectroscopy is a non-invasive technique that can stably record the neural response of the olfactory brain according to odors. Therefore, this study proposes a method for classifying odorants from data recorded neural responses of rat olfactory bulbs through near-infrared spectroscopy. Convolutional Neural Networks, one of the deep learning-based image classification technologies, was used as a classifier to classify three odor including air, isoamyl acetate, 2-heptanone. The classification result showed an average performance of 0.79 based on the F1-score.
화학적 초미세 발포 사출성형을 이용한 에어컨 드레인 펜의 공정 최적화에 대한 연구
김주권,곽재섭,김준민,이준한,김종선,Kim, Joo-Kwon,Kwak, Jae-Seob,Kim, Jun-Min,Lee, Jun-Han,Kim, Jong-Sun 한국금형공학회 2017 한국금형공학회지 Vol.11 No.2
In this study, we applied microcellular foaming injection molding process to improve the performance of system air-conditioner drain fan which had been produced by injection molding process and studied the optimization of process conditions through 6-sigma process and response surface method (RSM) to reduce weight and deformation of products. Additive type, melt temperature, mold temperature, and injection screw shape were selected as the factor affecting the weight and deformation of the products by carrying out analysis of trivial many through ANOVA and design of experiment (DOE) method. Among the effect factor, we set the addictive type to Long G/F and screw shape to foaming screw which had the highest level of weight reduction and deformation reduction. The amount of foaming agent gas was set at 60 ml, which was the limit beyond which the weight of product did not decrease any more. For melt temperature and mold temperature, we studied the conditions where both weight and deformation were minimized using the RSM. As a result, we set the melt temperature to $250^{\circ}C$, fixed mold temperature to $20^{\circ}C$, and moving mold temperature to $40^{\circ}C$. The improvement effect was analyzed by appling the selected optimal conditions to the production process using the microcellular foaming injection molding. The results showed that the mean weight of product was measured to be 1,420g which was 19% lower than that measured in the current process. The standard deviations of the weights were found to be similar to those in the current process and it showed a low dispersion. The mean deformation was measured to be 0.9237mm, which represented a 57% reduction compared to the mean deformation in the current process, and the standard deviation decreased from 0.3298mm to 0.1398mm. Moreover, we analyzed the process capability for deformation, and the results showed that the short-term process capability increased from 2.73 to 6.60 which was even higher than targeted level of 6.0.
VR 콘텐츠를 이용한 신체검사 개발 및 사용자 만족도 연구
안호원(Ho-Won An),김준민(Jun-Min Kim) 한국콘텐츠학회 2021 한국콘텐츠학회논문지 Vol.21 No.6
본 연구는 고령화 및 저출산에 따라 보건의료분야의 만성질환의 증가 및 전문 인력부족 등의 문제를 해결하고 효율적인 헬스케어를 제공하고자 VR 콘텐츠를 이용한 신체검사의 효과성을 연구하였다. 이에 본 연구는 HTC VIVE Pro VR과 스틱 컨트롤러를 착용하여 원스톱 VR 콘텐츠 신체검사 시스템을 구현하였다. 시스템은 1단계부터 5단계까지이며, 색각테스트, 기억력테스트, 청력테스트, 반응속도테스트, 순간인지능력테스트의 5단계를 순차적으로 거쳐 최종 신체 나이를 판별하는 솔루션으로 구성된다.또한 본 연구가 개발한 원스톱 VR 콘텐츠 신체검사 시스템에 대해 대전 소재 A 상급종합병원 건강검진센터 및 VR/AR 임상시험 센터에 내원한 정상인을 대상으로 사용자 만족도를 검증한 결과, 전체적인 만족도와 재이용의사가 높게 나타났고, 성별에 따라 5단계 순간인지테스트는 유의한 차이가 있었으며, 연령에 따라서 4단계 반응속도테스트, 5단계 순간인지테스트에서 유의한 차이가 나타났다. This study aims to the effectiveness of physical examination using VR contents to solve problems such as the increase in chronic diseases and shortage of professional manpower in the health care field according to the aging and low birth rate, and to provide efficient healthcare. Therefore, this study implemented a one-stop VR content physical examination system by wearing HTC VIVE Pro VR and a stick controller. The system is from step 1 to step 5, and the final body age is determined and a simple solution is provided through five steps sequentially: color blind test, memory test, audiogram test, reaction speed test, and instantaneous cognitive ability test. In addition, for the one-stop VR content physical examination system developed by this study, as a result of verifying the user satisfaction for normal people who visited the health examination center and VR/AR clinical trial center of certified tertiary hospital in Daejeon, the overall satisfaction and the intention to reuse Was high, and according to gender, there was a significant difference in the 5-step test, and according to the age, there were significant differences in the 4-step test and the 5-step test.
수술 중 신경의 시각화를 위한 시분할 액정-편광 이미징 시스템 개발 및 타당성 검증
이우람(Woo Ram Lee),임창균(Changkyun Im),임재관(Jaekwan Lim),김준민(Jun-Min Kim) 한국정보기술학회 2020 한국정보기술학회논문지 Vol.18 No.5
Since thin or buried nerves are difficult to recognize with the naked eye, many cases have been reported leading to surgical side effects or complications. Therefore, in this paper, we propose a liquid crystal-polarized light imaging(LCPi) system with a simple structure that allows visualization of nerves during surgery. The LCPi system is designed to acquire polarization images of 0 and 90 degrees by time division and to identify the region having polarization characteristics through image processing. The validation of the system was accomplished by imaging around the sciatic nerve of the rat and visualizing nerves against the surrounding tissue. The LCPi system is expected to be used as an intraoperative nerve visualization system to minimize nerve damage after the optimization process.